What problem does it solve? Before fixing a bug or building a feature in an unfamiliar repository, engineers often guess the package manager, framework, or run commands — leading to corrupted lockfiles, wrong test runs, and fixes that never address the real failure. This Skill grounds every repair plan in evidence gathered from the actual files, with read-only inspection and secret redaction. ## Core Features & Use Cases - Environment Detection: Identifies the package manager from lockfiles, confirms runtime versions against pins like .nvmrc, and maps the real dev/build/test script surface. - Failure Reproduction: Captures the exact failing command, full error output, and exit code before any edit is proposed. - Structural Mapping: Includes an absorbed codebase-cartography reference for tracing module layout, entry points, data flow, and high-risk files in sprawling codebases. - Use Case: You inherit a monorepo with no onboarding docs and a reported build failure. Run this Skill to confirm it uses pnpm workspaces, locate the owning package, list required env vars (values redacted), and reproduce the failure with its exit code before touching code. ## Quick Start Ask the AI to inspect this repository and report the package manager, scripts, framework, entrypoints, env variables, and reproduce the reported failure before making any changes.
